Intercellular communication mediated by extracellular vesicles (EVs) has recently attracted significant attention, and may contribute to diverse physiological processes including immune regulation and cancer metastasis. EVs are comprised of two major classes based upon their mechanism of biogenesis: microvesicles are shed directly from the plasma membrane, while exosomes are formed by intra-lumenal budding during multi-vesicular endosome (MVE) biogenesis (Colombo et al., 2014; van Niel et al., 2018) (Figure 1A). EVs contain a wide range of molecules, including antigens, signaling proteins, lipids, and microRNAs (miRNAs) that appear to induce significant biological changes in recipient cells. EV miRNAs released by both tumor and mesenchymal cells are of particular interest, as they may play important roles in metastasis (Becker et al., 2016). EV miRNAs secreted by hypothalamic stem cells may regulate aging in mice (Zhang et al., 2017). Moreover, EVs have attracted a great deal of attention due to their potential use in diagnostics and therapeutics (Becker et al., 2016). Despite their importance, how miRNAs are specifically packaged and released via the EV pathway is poorly understood. The biological significance of EVs also remains controversial, in part due to a lack of understanding of the molecular mechanisms that regulate their formation and release. Here we describe the development of a novel CRISPR screening platform using artificially barcoded miRNAs (bEXOmiRs) that we applied systematically to identify genes involved in EV biology. We show that EV release is regulated tightly by Glycogen Synthase Kinase three and Wnt signaling, and report new players in MVE exocytosis.

Figure 1 with 2 supplements see all

Download asset Open asset

Cell culture, antibodies and immunofluorescence--HEK293T cells were grown in Dulbecco's modified Eagle's media (DMEM) containing 10% fetal bovine serum, 2 mM L-glutamine, and penicillin (100 U/ml)/streptomycin (100 μg/ml). Wild type (WT) and Cas9-expressing (Morgens et al., 2017) K562 cells were grown in RPMI 1640 Medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 1 mM sodium pyruvate and penicillin (100 U/ml)/streptomycin (100 μg/ml). All cell lines were cultured at 37°C with 5% CO₂; cells were routinely tested for Mycoplasma using either a MycoAlert Mycoplasma Detection kit (Lonza LT07-318) or PCR. Their identity was assumed to be determined by ATCC. Extracellular vesicle-free media was prepared by overnight ultracentrifugation of RPMI supplemented with 20% FBS at 100,000 X g in a 45Ti rotor, as described (Théry et al., 2006).

Transfections of HEK293T cells plated in 10 cm dishes were carried out using 1.5 μg of bEXOmiR-encoding plasmid DNA (pEGFP-bEXOmiR) mixed with 1 ml serum-free DMEM and 30 μL polyethylenimine (PEI, 1 mg/ml, pH 7.0, PolySciences Inc.) and incubated at room temperature for 30 min. The mixture was added to cells that were grown for 48 hr before EV isolation. For Bafilomycin-A1 (Sigma-Aldrich) treatment, the drug was added to cell media at 10 nM and cells were cultured for 20 hr before EV collection.

Mouse monoclonal anti-α-Tubulin was from Sigma-Aldrich, rabbit polyclonal anti-ARHGEF18 was from GeneTex and anti-Rab27A was from Synaptic Systems. Mouse monoclonal anti-CD81, anti-β-Catenin and anti-Golgin 97 were from BD biosciences. Polyclonal anti-GDIβ was previously described (Dirac-Svejstrup et al., 1994). Rabbit polyclonal anti-Flotillin-1 antibody was from Novus Biologicals (NBP1-79022). Polyclonal anti-Calnexin and monoclonal anti-HSC70 antibodies were from Santa Cruz. Mouse monoclonal anti-LAMP1 antibody culture supernatant was from Developmental Studies Hybridoma Bank (University of Iowa, Iowa city, IA). Mouse monoclonal anti-LBPA and anti-CD63 antibodies were from EMD Millipore (MABT837) and DHSB (H5C6), respectively.

Immunofluorescence-- HeLa cells were transfected with fluorescently labeled siRNAs (DY547 label; Dharmacon) using DharmaFECT one transfection reagent, cultured for 48 hr, and replated on glass coverslips. To visualize K562 cells, cells were attached to glass coverslips using cytospin (Shandon) at 800 rpm for 5 min. Cells were then fixed with 3.7% (v/v) paraformaldehyde for 15 min, washed and permeabilized/blocked with 0.1% saponin/1% BSA-PBS before staining with anti-LAMP1 antibody followed by goat anti-rabbit Alexa 488 (1:2,000). Nuclei were stained using 0.1 µg/ml DAPI (Sigma-Aldrich). Coverslips were mounted on glass slides with Mowiol and imaged using a spinning disk confocal microscope (Yokogawa) with an electron multiplying charge-coupled device (EMCCD) camera (Andor, UK) and a 100 × 1.4 NA oil immersion objective. Finally, unbiased quantification of CD63-positive vesicles was performed using CellProfiler software (Carpenter et al., 2006).

Immunoblotting--HEK293T or K562 cells were harvested after 24–48 hr and lysed in lysis buffer (50 mM Hepes, pH 7.4, 150 mM NaCl, 5 mM MgCl2, and 1% Triton X-100) supplemented with protease inhibitors. After centrifugation at 12,000 g for 10 min, protein concentrations were measured in the cleared lysates. Equal amounts were resolved by SDS-PAGE followed by immunoblot analysis. EV final pellets were resuspended in SDS-PAGE loading dye (50 mM Tris-HCl pH 6.8, 6% glycerol, 2% SDS, 0.2% bromophenol blue), heated briefly at 95°C and resolved by 10% SDS-PAGE. After transfer to nitrocellulose and antibody incubation, blots were detected with ECL western blotting detection substrate or visualized using LI-COR Odyssey Imaging System. Isolation of cytosolic fractions from cell post-nuclear supernatants was performed as described (Lu and Pfeffer, 2013). For optimal detection of EV fraction markers (such as CD81), no reducing agent was added to the SDS-PAGE loading dye. Immunoblot signals visualized using LI-COR were analysed using ImageJ and presented as average ±SEM.

EV isolation--K562 suspension cell cultures were seeded at 5 × 10⁵ cells/ml in EV-free RPMI and grown for 24–48 hr. Cell cultures were centrifuged at 300 × g, 5 min to pellet cells, and supernatants were centrifuged again at 2100 × g, 20 min to pellet dead cells. The resulting supernatant was spun again at 10,000 × g, 30 min to pellet cell debris and large apoptotic bodies; for large scale screen cultures this step was replaced by filtration through 0.45 µm filter units (Thermo Scientific). After this, filtration through 0.22 µm filter units (Thermo Scientific) was performed. Filtered media were ultracentrifuged for 70 min at 100,000 × g in 45Ti rotor to pellet extracellular vesicles. The supernatant was discarded and the pellet re-suspended in phosphate-buffered saline (PBS) and centrifuged again for 70 min at 100,000 × g in TLA100.2 rotor. Discontinuous sucrose gradient flotation was as described (Shurtleff et al., 2016). EV pellets were frozen in liquid nitrogen and stored at −80 ˚C prior to further processing.

Nanoparticle tracking analysis--EVs were purified from control and CRISPR KO K562 polyclonal cell cultures as described (see EV isolation). After final ultracentrifugation, pelleted EVs were resuspended in PBS and analyzed on a NanoSight NS300 (Malvern Panalytical) using a sCMOS camera. Three, 60 s 25 Hz movies were taken per sample. Data were analyzed using NTA 3.1 software (Malvern Panalytical).

Flow cytometry analysis of LAMP1--K562 cells were treated with either DMSO or CHIR99021. After 24 h cells were harvested and washed before fixation in 2% paraformaldehyde, permeabilization and staining with anti-LAMP1 and Alexa Fluor 488 goat anti–mouse antibodies (Invitrogen). Quantification of LAMP1 total fluorescence was performed using a BD Accuri C6 flow cytometer and data were analyzed using Flowjo software (Tree Star).

Total internal reflection fluorescence (TIRF) microscopy--HeLa cells stably expressing pHluorin-CD63 were transfected with fluorescently labeled siRNAs (DY547 label; Dharmacon) using DharmaFECT one transfection reagent, cultured for 48 hr, and replated on chambered coverglass (Thermo Scientific). After 24 hr, cells containing fluorescent siRNAs were imaged on an inverted microscope (Nikon Eclipse Ti) equipped with an electron-multiplying charge-coupled device camera (ANDOR iXon^EM+), an Apo TIRF 100X/1.49 oil DIC objective, and Nikon Perfect Focus System. Time lapse videos were acquired at 3 Hz with Nikon NIS Elements software. All imaging experiments were performed at 37°C in a humidified incubator (5% CO₂). Fusion events for each cell were detected and quantified as sudden increases in fluorescence above the plasma membrane background as described (Verweij et al., 2018). At least 20 cells were imaged per condition in two independent experiments.

bEXOmiR design--Initial development of bEXOmiRs was based on previous artificial shRNAs designs (Chang et al., 2013) incorporating endogenous sequences derived from the putative exosomal microRNA hsa-miR-601. Specifically, a seven nt motif (GGAGGAG, ‘EXO motif’) together with a contiguous loop sequence (both derived from hsa-miR-601) were placed adjacent to a 15 nt random sequence (barcode) with perfect base complementarity, positioned at the base of the miRNA stem region (Figure 1—figure supplement 1A). The resulting precursor hairpin was flanked by hsa-miR-30a context sequences. For the initial bEXOmiR test library, a custom C-script was written to design barcode sequences based upon specific rules: 40–60% GC content, minimum Hamming distance of 4 or more between any two barcodes, and absence of homopolymers (AAAA, UUUU, CCCC, etc.) and restriction enzyme cut sites used for cloning purposes. With this strategy, 2500 bEXOmiRs were designed, and two versions of each were used, one with an additional A-C mismatch added to ensure proper primary-miR processing (Chang et al., 2013) for a total of 5000 bEXOmiRs.

To generate ~25,000 bEXOmiRs used in the MMM screen sublibrary, 15 base-pair barcodes were similarly generated with a minimum Hamming distance of four, moderate GC content (40–60%), and absence of homopolyers (GGGG, AAAA etc) and restriction sites. The first base-pair was altered to create a mismatch at the base of the hairpin to ensure proper processing (either A-C, T-C, G-A, or C-A). For subsequent sublibraries, this ~25,000 bEXOmiR set was filtered for either bEXOmiRs that failed to be detected in WT exosome samples (counts less than 10) or were detected at such high representation in WT exosomes as to interfere with sequencing depth (count >10,000). Additional barcodes were randomly generated as above for larger libraries.

Stem-loop RT-PCR and qRT-PCR--For bEXOmiR and endogenous miRNA detection in cellular or EV fractions, RNA was extracted from intact HEK293T or K562 cells, or isolated EVs using TRIzol, according to the manufacturer. EV RNAs (200–500 ng) were reverse transcribed into cDNA using Stem-Loop primers and SuperScript III FirstStrand Synthesis System (Life Technologies) according to a pulsed RT protocol (Varkonyi-Gasic and Hellens, 2011). Stem-loop RT primers are summarized in Supplementary file 3. qRT-PCR experiments used a qPCR mix recipe: 1X Phusion HF buffer (NEB), 200 µM dNTPs, 1X SYBR green I (Invitrogen), HighFidelity DNA Polymerase (M0530S, NEB), and 100 nM each primer. The following programs were used: (98°C, 3’, then 40X cycles of 98°C 15 s, 60°C 20 s, 72°C 3 s). A synthetic spike-in RT RNA (Supplementary file 3) was added to each EV sample to control for variability in the starting material and the downstream RNA isolation step. The sequences of PCR primers are summarized in Supplementary file 3. qPCR was run in a BioRad CFX96 Real-Time System. PCR amplification products were analyzed on 4% agarose gels stained with ethidium bromide. For qPCR experiments related to Figure 7, total cell RNA was reverse transcribed into cDNA using oligo(dT)20 primers and SuperScript III FirstStrand Synthesis System (Life Technologies). qPCR was performed using iTaq Universal SYBR Green Supermix (BioRad) using gene-specific primers (Supplementary file 3) and run in a ViiA 7 Real-Time PCR System (Thermo Fisher Scientific). Transcript levels relative to HPRT1 were calculated using the ∆Ct method.

RNase protection assay--K562 cells expressing bEXOmiRs were grown for 24 hr in RPMI and EVs were isolated from conditioned media. EVs were then treated with RNase A (10 µg/ml in PBS)±Triton X-100 (0.25%), or with PBS alone, for 30 min at room temperature. RNase inhibitor (150 U/ml, Thermo Fisher Scientific) was then added and EV-RNA isolated with TRIzol reagent (Life Technologies).

Construction of sgRNA-bEXOmiR plasmid libraries--Oligonucleotides encoding sgRNAs (Morgens et al., 2017) coupled to individual bEXOmiRs were synthesized as pooled libraries (Agilent). BlpI and EcoRI restriction sites were introduced between sgRNA and bEXOmiR sequences. Focused sublibraries were PCR-amplified with primer sequences common to each sublibrary, digested with BstXI and MfeI enzymes, ligated into BstXI/EcoRI-digested pMCB306 vector, and transformed into electro-competent cells (Lucigen). Plasmid DNA was isolated using a QIAGEN kit. Pooled plasmids were then digested with BlpI and EcoRI, and an expression cassette encoding a puromycin-resistance gene and a GFP-ORF separated by a T2A sequence under EF-1 alpha promoter, cut with the same restriction enzyme pair, was inserted between sgRNA and bEXOmiR sequences. This generated nine focused sublibraries (48, Supplementary file 4), which together contained ~210,000 elements targeting ~20,000 genes with ~13,000 non-targeting and safe-targeting controls (Morgens et al., 2017); 10 sgRNA per gene, each sgRNA associated to a unique bEXOmiR).

Lentivirus production and cell infection--Lentivirus production and infection were performed as described (Kampmann et al., 2014). Briefly, HEK293T cells were co-transfected with bEXOmiR libraries or individual sgRNA vectors and third generation packaging plasmids (pVSV-G, pMDL, pRSV) using polyethylenimine. Lentivirus was harvested after 48 h-72 h and filtered through a 0.45 μm filter (Millipore). A Spinfection method was used to deliver sgRNA-bEXOmiR lentiviral plasmid sublibraries into K562 cells. Briefly, K562 cells were re-suspended in lentiviral supernatant supplemented with 8 µg/ml polybrene, distributed into 6-well plates, and centrifuged for 2 hr at 1000 × g, 33°C. Lentiviral supernatant was removed and cells were grown in fresh RPMI. After 48 hr recovery, 1.75 µg/ml of puromycin was added to select the cells for stable integration of the virus. Selection was monitored by flow cytometry and puromycin was removed from the media once >95% of the cells were GFP-positive.

CRISPR/Cas9-bEXOmiR screens in K562 cells--For genome-wide coverage, a total of nine, focused CRISPR/Cas9 deletion screens using nine different sublibraries of sgRNA-bEXOmiR associations (Supplementary file 4) were performed. WT and Cas9-expressing K562 cells were infected at low MOI (<1). Transduced cells were selected and expanded in puromycin-supplemented media over 5–7 days before conducting experiments at 8–10 days post-infection. Initially, we performed two independent pilot screens focused on membrane trafficking, mitochondrial and motility genes (MMM screen). Subsequent screens were conducted in duplicate to further minimize technical variation observed in previous screens. Screens were started at 10,000–15,000 fold sublibrary coverage (elements per ml of cell culture) to ensure maximal bEXOmiR representation in EVs but also to minimize variability due to downstream loss of barcodes during RNA isolation and fractionation. Isolation of EVs was performed 48 hr later using protocol P2 (Figure 1—figure supplement 1B). Cellular fractions with at least 1000-fold sublibrary coverage were collected at the start (0 hr) and end (48 hr) of each screen.

bEXOmiR isolation and purification from screen EV RNA--Total RNA was extracted from each EV sample using TRIzol reagent (Life Technologies), as described (Lässer, 2013). The resulting RNA pellet was re-suspended in nuclease-free water and equal amount of 2x RNA loading dye was added (90% formamide, 50 mM HEPES, bromophenol blue, xylene cyanide). Fractionation of total RNA was carried out on denaturating polyacrylamide gels containing 7% urea/15% acrylamide in Tris-borate-EDTA buffer (TBE). Custom 21 and 24 bp single-stranded RNA markers were used to indicate approximate miRNA migration on a gel from which gel band excision was monitored by UV shadowing. miRNAs were then gel-purified, reconstituted in nuclease-free water, and stored in −80 ˚C.

Small RNA library preparation and deep sequencing--bEXOmiR sequencing libraries from isolated EV miRNA was prepared using a modified protocol from TruSeq Small RNA Library Preparation Kit (Illumina). For each library,~1 µg of fractionated small RNA was ligated first to a 3’-adapter and then to 5’-adapter, followed by reverse transcription with custom primer (Supplementary file 3) containing a fragment complementary to the EXO motif in bEXOmiRs (to specifically enrich for bEXOmiRs in the final sequencing library). Resulting cDNA was then uniquely indexed by PCR to generate the final sequencing library. Initial sequencing trials showed some rRNA contaminant fragments (specially 45S5) that were predominant in our sequencing data, which compromised sequencing depth. To remove rRNA 45S5, a specific hairpin oligo blocker (Supplementary file 3) containing a sequence complementary to the predominant rRNA fragment was incubated after ligation of the 3’ adapter to the miRNA (Roberts et al., 2015). This oligo blocked subsequent ligation of the 5’ adapter to the contaminant rRNA fragment, thus excluding it from later steps. After PCR amplification, indexed samples were separated on a 4.5% low melting point agarose gel and products corresponding to 140 bp were purified using a MinElute PCR Purification Kit (Qiagen). RNA concentration was determined by Qubit fluorometer (Thermo Fisher Scientific). Pooled libraries were denatured with NaOH and diluted to 2.3 pM according to the Illumina protocol, and sequenced using a NextSeq 500/550 High Output v2 kit on NextSeq 500 system (Illumina).

Sequencing data analysis--Sequencing of sgRNAs from DNA was trimmed and aligned using Bowtie version 1.1.2 with zero mismatches tolerated (Morgens et al., 2016). All counts from multi-mapping reads were used. For sequencing of barcodes from RNA, reads were trimmed to 14 base-pairs and aligned to the libraries using Bowtie version 1.1.2 with two mismatches tolerated and all multi-mapping reads used. Supplementary file 5 and 6 contain bEXOmiR and sgRNA sequencing counts derived from each duplicated screen. Gene-level effects and an associated log-likelihood ratio (confidence score) were then calculated with casTLE as previously described (Morgens et al., 2016). For each gene, casTLE combines measurements from 10 different bEXOmiRs (associated with the sgRNAs targeting each gene) and gives an effect size estimate for the gene perturbation (gene K.O.) as well as a p-value associated with that effect. casTLE estimates the maximum possible phenotype such that targeting elements are most likely to be found between this phenotype and zero. Additionally, casTLE allows data from multiple screen types or from replicates of the same screen type to be compared and combined by finding a single effect size consistent with all data (Morgens et al., 2016). Briefly, low count (<10) bEXOmiRs were filtered, and a median-normalized log count ratio was calculated for each barcode. Distributions of barcode enrichments corresponding to each gene were compared to the distribution of nontargeting and safe-targeting sgRNAs. P-values were then calculated by permuting bEXOmiRs corresponding to targeting sgRNAs. Code is available at https://bitbucket.org/dmorgens/castle (Copy archived at https://github.com/elifesciences-publications/dmorgens-castle).

Nanostring profiling was performed with the nCounter Human v3 microRNA expression assay. Control, SMPD3 and ARHGEF18 CRISPR-KO K562 polyclonal cell lines were seeded at 2.5 × 10⁵ cells/ml in EV-free RPMI medium and grown for 48 hr before EV isolation. RNA from cellular and exosomal fractions was extracted using Trizol as described (Lässer, 2013). EV RNA was resuspended in 4 µL H₂O and 3 µL (~300 ng) from each sample were used in the assay; 200 ng cellular RNA was used. RNA integrity was confirmed by Bioanalyzer before conducting the assay. Around 20 amol of synthetic spike-in RNA oligo (ath-miR-159a, see Supplementary file 3) was added to each EV sample prior to RNA extraction and used as a normalization factor in downstream EV miRNA data analysis. Raw read data from the nCounter microRNA assay were analyzed as described in the nCounter Expression Data Analysis Guide (NanoString Technologies). Data normalization for cellular miRs was perfomed using the geometric mean of the top 100 expressed miRNAs. Only miRs that passed signal-to-noise cut-off values and that were consistently detected in either cellular or EV fractions (274 and 74 miRNAs, respectively) throughout all samples, were included in the analysis and plotted using R software. Normalized miRNA counts (Supplementary file 7) from SMPD3 KO or ARHGEF18 KO samples were divided by control miRNA counts from independent experiments (n = 2) and plotted as percent of control miRNA abundance.

Generation of CRISPR-KO cell lines for validation experiments--Polyclonal, CRISPR-KO cell lines were generated by transducing K562 stably expressing SFFV-Cas9-BFP with Lentiviral vectors encoding hit-targeting sgRNAs; at least two different sgRNAs per hit were tested. Cultures were incubated for at least 7–10 d with the sgRNAs to ensure adequate depletion of target-protein levels before conducting experiments. TIDE analysis (https://tide.nki.nl/) was performed to assess CRISPR deletions. Briefly, genomic DNA was extracted from cells using a QIAamp DNA mini kit (Qiagen). A 500–700 bp region containing the sgRNA cut site was amplified by PCR using GoTaq Polymerase (Promega). PCR amplification products were separated on a 0.75% TAE-agarose gel, from which amplicon bands were excised, and DNA was purified from the gel slice using a QIAquick Gel Extraction kit (Qiagen). Sanger sequencing of PCR products followed by trace decomposition analysis with TIDE software (https://tide.nki.nl/) was used to determine indel frequency in CRISPR-KO cell lines compared with control cells expressing a control sgRNA (targeting a genomic safe-harbor region). For validation experiments, sgRNA-expressing cell lines that showed higher degree of knockout efficiency as determined by either TIDE and/or immunoblot analysis were used (Figure 5—figure supplement 1).

Wnt signaling--For Wnt3a stimulation, K562 cells were starved overnight in serum-free conditions. The next day, the medium was replaced with fresh EV-free RPMI supplemented with 100 ng/ml recombinant Wnt3a protein (R and D Systems), or Wnt3a combined with 200 ng/ml recombinant DKK1 protein (R and D Systems) and EVs were collected 24 hr later. The GSK3 specific inhibitor, CHIR99021 (Selleckchem) and LiCl were used at 10 µM and 10 mM, respectively for 24 hr before collection of EVs from the medium. Quantitation of immunoblots from each of these conditions was normalized to account for cell viability, as measured by flow cytometry (BD Accuri). C59 Porcupine inhibitor was the generous gift of Dr. Roel Nusse, Stanford University.

Functional Network analysis--Top hits were collectively queried in STRING (https://string-db.org/) to search for gene interactions according to co-expression, experimental or curated database sources. The resulting STRING file reporting pair-wise gene interactions was imported into Cytoscape software (Shannon et al., 2003) to create a visually comprehensive functional interactome. Interactions were represented as edges whose thickness was proportional to a calculated combined score derived from STRING analysis. The final Cytoscape map was further manually curated to include additional screen hits (nodes) that despite not being reported to interact, could still be clustered in a common functional category revealed in the screen.

All data generated or analysed during this study are included in the manuscript and supporting files.

1. 1. Arora S
   2. Saarloos I
   3. Kooistra R
   4. van de Bospoort R
   5. Verhage M
   6. Toonen RF

   (2017) SNAP-25 gene family members differentially support secretory vesicle fusion

   Journal of Cell Science 130:1877–1889.

   https://doi.org/10.1242/jcs.201889

   • PubMed
   • Google Scholar
2. 1. Artym VV
   2. Swatkoski S
   3. Matsumoto K
   4. Campbell CB
   5. Petrie RJ
   6. Dimitriadis EK
   7. Li X
   8. Mueller SC
   9. Bugge TH
   10. Gucek M
   11. Yamada KM

   (2015) Dense fibrillar collagen is a potent inducer of invadopodia via a specific signaling network

   The Journal of Cell Biology 208:331–350.

   https://doi.org/10.1083/jcb.201405099

   • PubMed
   • Google Scholar
3. 1. Bailey TL
   2. Johnson J
   3. Grant CE
   4. Noble WS

   (2015) The MEME Suite

   Nucleic Acids Research 43:W39–W49.

   https://doi.org/10.1093/nar/gkv416

   • PubMed
   • Google Scholar
4. 1. Becker A
   2. Thakur BK
   3. Weiss JM
   4. Kim HS
   5. Peinado H
   6. Lyden D

   (2016) Extracellular vesicles in cancer: cell-to-cell mediators of metastasis

   Cancer Cell 30:836–848.

   https://doi.org/10.1016/j.ccell.2016.10.009

   • PubMed
   • Google Scholar
5. 1. Carpenter AE
   2. Jones TR
   3. Lamprecht MR
   4. Clarke C
   5. Kang IH
   6. Friman O
   7. Guertin DA
   8. Chang JH
   9. Lindquist RA
   10. Moffat J
   11. Golland P
   12. Sabatini DM

   (2006) CellProfiler: image analysis software for identifying and quantifying cell phenotypes

   Genome Biology 7:R100.

   https://doi.org/10.1186/gb-2006-7-10-r100

   • PubMed
   • Google Scholar
6. 1. Cha DJ
   2. Franklin JL
   3. Dou Y
   4. Liu Q
   5. Higginbotham JN
   6. Demory Beckler M
   7. Weaver AM
   8. Vickers K
   9. Prasad N
   10. Levy S
   11. Zhang B
   12. Coffey RJ
   13. Patton JG

   (2015) KRAS-dependent sorting of miRNA to exosomes

   eLife 4:e07197.

   https://doi.org/10.7554/eLife.07197

   • PubMed
   • Google Scholar
7. 1. Chang K
   2. Marran K
   3. Valentine A
   4. Hannon GJ

   (2013) Creating an miR30-based shRNA vector

   Cold Spring Harbor Protocols 2013:631–635.

   https://doi.org/10.1101/pdb.prot075853

   • PubMed
   • Google Scholar
8. 1. Chiaverini C
   2. Beuret L
   3. Flori E
   4. Busca R
   5. Abbe P
   6. Bille K
   7. Bahadoran P
   8. Ortonne JP
   9. Bertolotto C
   10. Ballotti R

   (2008) Microphthalmia-associated transcription factor regulates RAB27A gene expression and controls melanosome transport

   Journal of Biological Chemistry 283:12635–12642.

   https://doi.org/10.1074/jbc.M800130200

   • PubMed
   • Google Scholar
9. 1. Colombo M
   2. Raposo G
   3. Théry C

   (2014) Biogenesis, secretion, and intercellular interactions of exosomes and other extracellular vesicles

   Annual Review of Cell and Developmental Biology 30:255–289.

   https://doi.org/10.1146/annurev-cellbio-101512-122326

   • PubMed
   • Google Scholar
10. 1. Conradt B
    2. Haas A
    3. Wickner W

    (1994) Determination of four biochemically distinct, sequential stages during vacuole inheritance in vitro

    The Journal of Cell Biology 126:99–110.

    https://doi.org/10.1083/jcb.126.1.99

    • PubMed
    • Google Scholar
11. 1. Cruciat CM
    2. Ohkawara B
    3. Acebron SP
    4. Karaulanov E
    5. Reinhard C
    6. Ingelfinger D
    7. Boutros M
    8. Niehrs C

    (2010) Requirement of prorenin receptor and vacuolar H+-ATPase-mediated acidification for Wnt signaling

    Science 327:459–463.

    https://doi.org/10.1126/science.1179802

    • PubMed
    • Google Scholar
12. 1. Dattilo V
    2. D'Antona L
    3. Talarico C
    4. Capula M
    5. Catalogna G
    6. Iuliano R
    7. Schenone S
    8. Roperto S
    9. Bianco C
    10. Perrotti N
    11. Amato R

    (2017) SGK1 affects RAN/RANBP1/RANGAP1 via SP1 to play a critical role in pre-miRNA nuclear export: a new route of epigenomic regulation

    Scientific Reports 7:45361.

    https://doi.org/10.1038/srep45361

    • PubMed
    • Google Scholar
13. 1. Dirac-Svejstrup AB
    2. Soldati T
    3. Shapiro AD
    4. Pfeffer SR

    (1994)

    Rab-GDI presents functional Rab9 to the intracellular transport machinery and contributes selectivity to Rab9 membrane recruitment

    The Journal of Biological Chemistry 269:15427–15430.

    • PubMed
    • Google Scholar
14. 1. Edgar JR
    2. Manna PT
    3. Nishimura S
    4. Banting G
    5. Robinson MS

    (2016) Tetherin is an exosomal tether

    eLife 5:e17180.

    https://doi.org/10.7554/eLife.17180

    • PubMed
    • Google Scholar
15. 1. Faure S
    2. Fort P

    (2011) Atypical RhoV and RhoU GTPases control development of the neural crest

    Small GTPases 2:310–313.

    https://doi.org/10.4161/sgtp.18086

    • PubMed
    • Google Scholar
16. 1. Gao J
    2. Takeuchi H
    3. Zhang Z
    4. Fukuda M
    5. Hirata M

    (2012) Phospholipase C-related but catalytically inactive protein (PRIP) modulates synaptosomal-associated protein 25 (SNAP-25) phosphorylation and exocytosis

    Journal of Biological Chemistry 287:10565–10578.

    https://doi.org/10.1074/jbc.M111.294645

    • PubMed
    • Google Scholar
17. 1. Glebov K
    2. Löchner M
    3. Jabs R
    4. Lau T
    5. Merkel O
    6. Schloss P
    7. Steinhäuser C
    8. Walter J

    (2015) Serotonin stimulates secretion of exosomes from microglia cells

    Glia 63:626–634.

    https://doi.org/10.1002/glia.22772

    • PubMed
    • Google Scholar
18. 1. Gross JC
    2. Chaudhary V
    3. Bartscherer K
    4. Boutros M

    (2012) Active Wnt proteins are secreted on exosomes

    Nature Cell Biology 14:1036–1045.

    https://doi.org/10.1038/ncb2574

    • PubMed
    • Google Scholar
19. 1. Hooper C
    2. Sainz-Fuertes R
    3. Lynham S
    4. Hye A
    5. Killick R
    6. Warley A
    7. Bolondi C
    8. Pocock J
    9. Lovestone S

    (2012) Wnt3a induces exosome secretion from primary cultured rat microglia

    BMC Neuroscience 13:144.

    https://doi.org/10.1186/1471-2202-13-144

    • PubMed
    • Google Scholar
20. 1. Hoshino D
    2. Kirkbride KC
    3. Costello K
    4. Clark ES
    5. Sinha S
    6. Grega-Larson N
    7. Tyska MJ
    8. Weaver AM

    (2013) Exosome secretion is enhanced by invadopodia and drives invasive behavior

    Cell Reports 5:1159–1168.

    https://doi.org/10.1016/j.celrep.2013.10.050

    • PubMed
    • Google Scholar
21. 1. Kampmann M
    2. Bassik MC
    3. Weissman JS

    (2014) Functional genomics platform for pooled screening and generation of mammalian genetic interaction maps

    Nature Protocols 9:1825–1847.

    https://doi.org/10.1038/nprot.2014.103

    • PubMed
    • Google Scholar
22. 1. Kosaka N
    2. Iguchi H
    3. Yoshioka Y
    4. Takeshita F
    5. Matsuki Y
    6. Ochiya T

    (2010) Secretory mechanisms and intercellular transfer of microRNAs in living cells

    Journal of Biological Chemistry 285:17442–17452.

    https://doi.org/10.1074/jbc.M110.107821

    • PubMed
    • Google Scholar
23. 1. Kosaka N
    2. Iguchi H
    3. Hagiwara K
    4. Yoshioka Y
    5. Takeshita F
    6. Ochiya T

    (2013) Neutral sphingomyelinase 2 (nSMase2)-dependent exosomal transfer of angiogenic microRNAs regulate cancer cell metastasis

    Journal of Biological Chemistry 288:10849–10859.

    https://doi.org/10.1074/jbc.M112.446831

    • PubMed
    • Google Scholar
24. 1. Lässer C

    (2013) Identification and analysis of circulating exosomal microRNA in human body fluids

    Methods in Molecular Biology 1024:109–128.

    https://doi.org/10.1007/978-1-62703-453-1_9

    • PubMed
    • Google Scholar
25. 1. Li B
    2. Antonyak MA
    3. Zhang J
    4. Cerione RA

    (2012) RhoA triggers a specific signaling pathway that generates transforming microvesicles in cancer cells

    Oncogene 31:4740–4749.

    https://doi.org/10.1038/onc.2011.636

    • PubMed
    • Google Scholar
26. 1. Lu A
    2. Pfeffer SR

    (2013) Golgi-associated RhoBTB3 targets cyclin E for ubiquitylation and promotes cell cycle progression

    The Journal of Cell Biology 203:233–250.

    https://doi.org/10.1083/jcb.201305158

    • PubMed
    • Google Scholar
27. 1. Medina DL
    2. Fraldi A
    3. Bouche V
    4. Annunziata F
    5. Mansueto G
    6. Spampanato C
    7. Puri C
    8. Pignata A
    9. Martina JA
    10. Sardiello M
    11. Palmieri M
    12. Polishchuk R
    13. Puertollano R
    14. Ballabio A

    (2011) Transcriptional activation of lysosomal exocytosis promotes cellular clearance

    Developmental Cell 21:421–430.

    https://doi.org/10.1016/j.devcel.2011.07.016

    • PubMed
    • Google Scholar
28. 1. Menck K
    2. Sönmezer C
    3. Worst TS
    4. Schulz M
    5. Dihazi GH
    6. Streit F
    7. Erdmann G
    8. Kling S
    9. Boutros M
    10. Binder C
    11. Gross JC

    (2017) Neutral sphingomyelinases control extracellular vesicles budding from the plasma membrane

    Journal of Extracellular Vesicles 6:1378056.

    https://doi.org/10.1080/20013078.2017.1378056

    • PubMed
    • Google Scholar
29. 1. Miao Y
    2. Li G
    3. Zhang X
    4. Xu H
    5. Abraham SN

    (2015) A TRP channel senses lysosome neutralization by pathogens to trigger their expulsion

    Cell 161:1306–1319.

    https://doi.org/10.1016/j.cell.2015.05.009

    • PubMed
    • Google Scholar
30. 1. Morgens DW
    2. Deans RM
    3. Li A
    4. Bassik MC

    (2016) Systematic comparison of CRISPR/Cas9 and RNAi screens for essential genes

    Nature Biotechnology 34:634–636.

    https://doi.org/10.1038/nbt.3567

    • PubMed
    • Google Scholar
31. 1. Morgens DW
    2. Wainberg M
    3. Boyle EA
    4. Ursu O
    5. Araya CL
    6. Tsui CK
    7. Haney MS
    8. Hess GT
    9. Han K
    10. Jeng EE
    11. Li A
    12. Snyder MP
    13. Greenleaf WJ
    14. Kundaje A
    15. Bassik MC

    (2017) Genome-scale measurement of off-target activity using Cas9 toxicity in high-throughput screens

    Nature Communications 8:15178.

    https://doi.org/10.1038/ncomms15178

    • PubMed
    • Google Scholar
32. 1. Nusse R
    2. Clevers H

    (2017) Wnt/β-catenin signaling, disease, and emerging therapeutic modalities

    Cell 169:985–999.

    https://doi.org/10.1016/j.cell.2017.05.016

    • PubMed
    • Google Scholar
33. 1. Ostrowski M
    2. Carmo NB
    3. Krumeich S
    4. Fanget I
    5. Raposo G
    6. Savina A
    7. Moita CF
    8. Schauer K
    9. Hume AN
    10. Freitas RP
    11. Goud B
    12. Benaroch P
    13. Hacohen N
    14. Fukuda M
    15. Desnos C
    16. Seabra MC
    17. Darchen F
    18. Amigorena S
    19. Moita LF
    20. Thery C

    (2010) Rab27a and Rab27b control different steps of the exosome secretion pathway

    Nature Cell Biology 12:19–30.

    https://doi.org/10.1038/ncb2000

    • PubMed
    • Google Scholar
34. 1. Özhan G
    2. Sezgin E
    3. Wehner D
    4. Pfister AS
    5. Kühl SJ
    6. Kagermeier-Schenk B
    7. Kühl M
    8. Schwille P
    9. Weidinger G

    (2013) Lypd6 enhances Wnt/β-catenin signaling by promoting Lrp6 phosphorylation in raft plasma membrane domains

    Developmental Cell 26:331–345.

    https://doi.org/10.1016/j.devcel.2013.07.020

    • PubMed
    • Google Scholar
35. 1. Peters C
    2. Andrews PD
    3. Stark MJ
    4. Cesaro-Tadic S
    5. Glatz A
    6. Podtelejnikov A
    7. Mann M
    8. Mayer A

    (1999) Control of the terminal step of intracellular membrane fusion by protein phosphatase 1

    Science 285:1084–1087.

    https://doi.org/10.1126/science.285.5430.1084

    • PubMed
    • Google Scholar
36. 1. Ploper D
    2. De Robertis EM

    (2015) The MITF family of transcription factors: Role in endolysosomal biogenesis, Wnt signaling, and oncogenesis

    Pharmacological Research 99:36–43.

    https://doi.org/10.1016/j.phrs.2015.04.006

    • PubMed
    • Google Scholar
37. 1. Ploper D
    2. Taelman VF
    3. Robert L
    4. Perez BS
    5. Titz B
    6. Chen HW
    7. Graeber TG
    8. von Euw E
    9. Ribas A
    10. De Robertis EM

    (2015) MITF drives endolysosomal biogenesis and potentiates Wnt signaling in melanoma cells

    PNAS 112:E420–E429.

    https://doi.org/10.1073/pnas.1424576112

    • PubMed
    • Google Scholar
38. 1. Roberts BS
    2. Hardigan AA
    3. Kirby MK
    4. Fitz-Gerald MB
    5. Wilcox CM
    6. Kimberly RP
    7. Myers RM

    (2015) Blocking of targeted microRNAs from next-generation sequencing libraries

    Nucleic Acids Research 61:gkv724.

    https://doi.org/10.1093/nar/gkv724

    • Google Scholar
39. 1. Santangelo L
    2. Giurato G
    3. Cicchini C
    4. Montaldo C
    5. Mancone C
    6. Tarallo R
    7. Battistelli C
    8. Alonzi T
    9. Weisz A
    10. Tripodi M

    (2016) The RNA-Binding Protein SYNCRIP Is a Component of the Hepatocyte Exosomal Machinery Controlling MicroRNA Sorting

    Cell Reports 17:799–808.

    https://doi.org/10.1016/j.celrep.2016.09.031

    • PubMed
    • Google Scholar
40. 1. Sercan Z
    2. Pehlivan M
    3. Sercan HO

    (2010) Expression profile of WNT, FZD and sFRP genes in human hematopoietic cells

    Leukemia Research 34:946–949.

    https://doi.org/10.1016/j.leukres.2010.02.009

    • PubMed
    • Google Scholar
41. 1. Shannon P
    2. Markiel A
    3. Ozier O
    4. Baliga NS
    5. Wang JT
    6. Ramage D
    7. Amin N
    8. Schwikowski B
    9. Ideker T

    (2003) Cytoscape: a software environment for integrated models of biomolecular interaction networks

    Genome Research 13:2498–2504.

    https://doi.org/10.1101/gr.1239303

    • PubMed
    • Google Scholar
42. 1. Shurtleff MJ
    2. Temoche-Diaz MM
    3. Karfilis KV
    4. Ri S
    5. Schekman R

    (2016) Y-box protein 1 is required to sort microRNAs into exosomes in cells and in a cell-free reaction

    eLife 5:e19276.

    https://doi.org/10.7554/eLife.19276

    • PubMed
    • Google Scholar
43. 1. Sim AT
    2. Baldwin ML
    3. Rostas JA
    4. Holst J
    5. Ludowyke RI

    (2003) The role of serine/threonine protein phosphatases in exocytosis

    Biochemical Journal 373:641–659.

    https://doi.org/10.1042/bj20030484

    • PubMed
    • Google Scholar
44. 1. Sinha S
    2. Hoshino D
    3. Hong NH
    4. Kirkbride KC
    5. Grega-Larson NE
    6. Seiki M
    7. Tyska MJ
    8. Weaver AM

    (2016) Cortactin promotes exosome secretion by controlling branched actin dynamics

    The Journal of Cell Biology 214:197–213.

    https://doi.org/10.1083/jcb.201601025

    • PubMed
    • Google Scholar
45. 1. Taelman VF
    2. Dobrowolski R
    3. Plouhinec JL
    4. Fuentealba LC
    5. Vorwald PP
    6. Gumper I
    7. Sabatini DD
    8. De Robertis EM

    (2010) Wnt signaling requires sequestration of glycogen synthase kinase 3 inside multivesicular endosomes

    Cell 143:1136–1148.

    https://doi.org/10.1016/j.cell.2010.11.034

    • PubMed
    • Google Scholar
46. 1. Terry SJ
    2. Zihni C
    3. Elbediwy A
    4. Vitiello E
    5. Leefa Chong San IV
    6. Balda MS
    7. Matter K

    (2011) Spatially restricted activation of RhoA signalling at epithelial junctions by p114RhoGEF drives junction formation and morphogenesis

    Nature Cell Biology 13:159–166.

    https://doi.org/10.1038/ncb2156

    • PubMed
    • Google Scholar
47. 1. Terry SJ
    2. Elbediwy A
    3. Zihni C
    4. Harris AR
    5. Bailly M
    6. Charras GT
    7. Balda MS
    8. Matter K

    (2012) Stimulation of cortical myosin phosphorylation by p114RhoGEF drives cell migration and tumor cell invasion

    PLoS ONE 7:e50188.

    https://doi.org/10.1371/journal.pone.0050188

    • PubMed
    • Google Scholar
48. 1. Théry C
    2. Amigorena S
    3. Raposo G
    4. Clayton A

    (2006) Isolation and characterization of exosomes from cell culture supernatants and biological fluids

    Current Protocols in Cell Biology 30:3.22.1–3.22.3.

    https://doi.org/10.1002/0471143030.cb0322s30

    • Google Scholar
49. 1. Treiber T
    2. Treiber N
    3. Plessmann U
    4. Harlander S
    5. Daiß JL
    6. Eichner N
    7. Lehmann G
    8. Schall K
    9. Urlaub H
    10. Meister G

    (2017) A Compendium of RNA-Binding Proteins that Regulate MicroRNA Biogenesis

    Molecular Cell 66:270–284.

    https://doi.org/10.1016/j.molcel.2017.03.014

    • PubMed
    • Google Scholar
50. 1. Tsuji T
    2. Ohta Y
    3. Kanno Y
    4. Hirose K
    5. Ohashi K
    6. Mizuno K

    (2010) Involvement of p114-RhoGEF and Lfc in Wnt-3a- and dishevelled-induced RhoA activation and neurite retraction in N1E-115 mouse neuroblastoma cells

    Molecular Biology of the Cell 21:3590–3600.

    https://doi.org/10.1091/mbc.e10-02-0095

    • PubMed
    • Google Scholar
51. 1. van Niel G
    2. D'Angelo G
    3. Raposo G

    (2018) Shedding light on the cell biology of extracellular vesicles

    Nature Reviews Molecular Cell Biology 19:213–228.

    https://doi.org/10.1038/nrm.2017.125

    • PubMed
    • Google Scholar
52. 1. Varkonyi-Gasic E
    2. Hellens RP

    (2011) Quantitative stem-loop RT-PCR for detection of microRNAs

    Methods in Molecular Biology 744:145–157.

    https://doi.org/10.1007/978-1-61779-123-9_10

    • PubMed
    • Google Scholar
53. 1. Verweij FJ
    2. Bebelman MP
    3. Jimenez CR
    4. Garcia-Vallejo JJ
    5. Janssen H
    6. Neefjes J
    7. Knol JC
    8. de Goeij-de Haas R
    9. Piersma SR
    10. Baglio SR
    11. Verhage M
    12. Middeldorp JM
    13. Zomer A
    14. van Rheenen J
    15. Coppolino MG
    16. Hurbain I
    17. Raposo G
    18. Smit MJ
    19. Toonen RFG
    20. van Niel G
    21. Pegtel DM

    (2018) Quantifying exosome secretion from single cells reveals a modulatory role for GPCR signaling

    The Journal of Cell Biology 217:1129–1142.

    https://doi.org/10.1083/jcb.201703206

    • PubMed
    • Google Scholar
54. 1. Villarroya-Beltri C
    2. Gutiérrez-Vázquez C
    3. Sánchez-Cabo F
    4. Pérez-Hernández D
    5. Vázquez J
    6. Martin-Cofreces N
    7. Martinez-Herrera DJ
    8. Pascual-Montano A
    9. Mittelbrunn M
    10. Sánchez-Madrid F

    (2013) Sumoylated hnRNPA2B1 controls the sorting of miRNAs into exosomes through binding to specific motifs

    Nature Communications 4:2980.

    https://doi.org/10.1038/ncomms3980

    • PubMed
    • Google Scholar
55. 1. Yi R
    2. Qin Y
    3. Macara IG
    4. Cullen BR

    (2003) Exportin-5 mediates the nuclear export of pre-microRNAs and short hairpin RNAs

    Genes & Development 17:3011–3016.

    https://doi.org/10.1101/gad.1158803

    • PubMed
    • Google Scholar
56. 1. Zaritsky A
    2. Tseng YY
    3. Rabadán MA
    4. Krishna S
    5. Overholtzer M
    6. Danuser G
    7. Hall A

    (2017) Diverse roles of guanine nucleotide exchange factors in regulating collective cell migration

    The Journal of Cell Biology 216:1543–1556.

    https://doi.org/10.1083/jcb.201609095

    • PubMed
    • Google Scholar
57. 1. Zhang Y
    2. Liu S
    3. Mickanin C
    4. Feng Y
    5. Charlat O
    6. Michaud GA
    7. Schirle M
    8. Shi X
    9. Hild M
    10. Bauer A
    11. Myer VE
    12. Finan PM
    13. Porter JA
    14. Huang SM
    15. Cong F

    (2011) RNF146 is a poly(ADP-ribose)-directed E3 ligase that regulates axin degradation and Wnt signalling

    Nature Cell Biology 13:623–629.

    https://doi.org/10.1038/ncb2222

    • PubMed
    • Google Scholar
58. 1. Zhang Y
    2. Kim MS
    3. Jia B
    4. Yan J
    5. Zuniga-Hertz JP
    6. Han C
    7. Cai D

    (2017) Hypothalamic stem cells control ageing speed partly through exosomal miRNAs

    Nature 548:52–57.

    https://doi.org/10.1038/nature23282

    • PubMed
    • Google Scholar

Thank you for sending your article entitled "Genome-wide interrogation of extracellular vesicle release using barcoded miRNAs reveals a role for Wnt signaling" for peer review at eLife. Your article is being evaluated by Anna Akhmanova as the Senior Editor, a Reviewing Editor, and three reviewers.

As you will see from the reviews, the reviewers' opinions of your study were mixed. All reviewers feel that the bEXOmiR experimental approach is clever and may hold promise for identifying EV biogenesis pathway components. However, while the reviewers acknowledge the potential trove of new information regarding EV biogenesis, they found the breadth and diversity of the 'hits' a matter of some concern, because for most cases it is not possible to distinguish "fundamental" versus "peripheral" hits in EV biogenesis pathways. It is a fair assumption that many of the hits impact EV biogenesis and release due to general perturbations to organelle function (that is, they do not directly participate in some aspect of EV biology). The consensus view is that additional experimentation is necessary to sufficiently validate the bEXOmiR approach and the screening data, and the reviewers request that you outline additional experimentation to address this. The reviewers would be especially encouraged by experimental data regarding mechanism: what aspect of EV biogenesis is affected by loss of ARHGEF18 and/or is regulated by Wnt signaling? The reviewers felt that the Wnt signaling and ARHGEF18 follow-up work was largely confirmatory of the original screen hits but did not provide strong additional evidence for a fundamental role of these factors in EV biogenesis or regulation. The reviewers would also have liked to see some additional data that would provide insight into the origin of the bEXOmiR-containing EV vesicle (plasma membrane? endosome? another organelle?).

Reviewer #1:

The manuscript describes the construction and validation of a genetic screening method that links specific sgRNA sequences with synthetic miRNAs that are secreted via extra-cellular vesicles (EVs). The method allows one to use deep sequencing to gauge the effects of targeted gene knockouts on EV-mediated release of miRNAs on a whole genome scale. The method was applied in replicate screens, and its effectiveness is suggested by the identification of a handful of gene products previously implicated in EV biogenesis and/or release. In addition, at least one gene, ARHGEF18, was shown to be required for EV-mediated release of endogenous miRNAs.

I find the method to be a clever implementation of miRNA and sgRNA processing pathways to link genetics with phenotype. The most important question is, does it work? The identification of gene products previously identified as required for EV release suggests that it does. The approach did, however, require the authors to tackle a few technical bugs (e.g., DNA sequencing, effects of different sgRNA-barcode combinations, etc.) and it is not clear if further, unknown biases (if any) impact the utility of the method as a general method for evaluating the genetics of EV biogenesis. A large number of genes not previously implicated in EV biology are asserted to play roles in EV biogenesis. At this stage, it is unclear if any of the 'new' hits play a direct role in EV cargo sorting, EV biogenesis, or EV release. As with most screens, the ultimate utility of the data will require substantial amount of follow-up work and will not be known for years to come.

Reviewer #2:

In this manuscript Lu and colleagues perform a genome wide screen to get insights into the biogenesis and regulation of Extracellular Vesicle release.

For this purpose, the authors set up an elegant system based on the generation of barcoded miRNAs that depict a sequence that has been previously shown to target miRNAs to multivesicular endosomes of T lymphocytes and in particular to their intralumenal vesicles (called exosomes once they are secreted). These findings from the group of Sanchez Madrid are largely exploited here (and would deserve better acknowledgement).

The barcoded miRNAs are then used as reporters of EV release which is followed by a genome wide screen coupled with CRISPR/Cas9.

The authors have performed an incredible amount of work, that certainly is a very interesting set up.

There are several concerns with the manuscript, interpretation of results and the claims. Finally, one is a little left with the question on how such approach can really answer fundamental questions and pinpoint general or particular/specific machineries required for EV biogenesis and release.

Essential revisions:

1) EVs comprise very different types of vesicles that are either endosome derived (so called exosomes) and plasma membrane derived. One important issue in the field is the possibility in the future to ascribe functions to the different subpopulations. Several studies have described and reported distinct machineries for the biogenesis and release of exosomes and microvesicles that can be cell type and cargo specific (summarized in different recent reviews from different authors). Therefore, it is unclear what is the approach that is here developed bringing to the complexity that has been observed and described for the very different systems.

2) It is unclear from the claims in the manuscript and from the experiments done if one is trying to get hints on the biogenesis and release of endosome-derived exosomes or plasma membrane derived vesicles. It would certainly be important for the general audience to be clearer on this issue that is a real concern.

With this in mind, the characterization made, that is far from being very extensive, would seem to indicate that the authors maybe rather focus on endosome derived exosomes.

This point should be consolidated better with additional proteins tested by Western blot, CD63, Tsg101. For example, CD63 should rather be used for the immunofluorescence experiments presented. Lamp1, if present in late endosomes, is much more enriched in lysosomes. The experiments presented in Figure 1 and Figure 6 are difficult then to interpret.

Lamp1 is far from being the best "exosome" marker as in most cells Lamp1 is on the limiting membrane of the organelle and it is Lamp2 that is rather on the intraluminal vesicles. Therefore, it would be important to control whether in the cells under study Lamp1 is targeted to the intraluminal vesicles (or even if this is a consequence of the barcoded miRNAs?)

3) Following point 2, electron microscopy combined with extra immunofluorescence imaging would be more then appropriate and required to be sure, even to set background for the all findings, that the barcoded miRNAs, once expressed are not impacting in cell/organelle morphology and/or if they are targeted to the right place. This could also help understanding the effects of Rab27A which in most cells where it is expressed (not all cells express the "a" isoform) is present in Golgi TGN. Rab27a is localized to Lysosome Related organelles only in specialized cells. Therefore, is Rab27a in these manipulated cells recruited to multivesicular endosomes that may specialize after introduction of the barcoded miRs? For example, this is the case in HeLa cells expressing MHC Class II and all associated molecules for antigen presentation (Ostrowski et al., 2010). The sequence used has been described to my knowledge only in T cells, is it really driving the miRs to the right place in very different cells? In conclusion extra controls would support part of the claims. This would then help understanding what subpopulation of EVs is targeted in the study.

4) I would be more careful generalizing the concept based on the two cell lines used in this study although both express wnt.

5) The manuscript describes the involvement of Wnt signaling in the regulation of « EV biogenesis ». For me it is unclear what is really the mechanism of Wnt action: is it on the biogenesis per se or on the release/secretion step which are different parts of the pathway? It could be only on the release but whether there are biogenetic functions remains unclear. Are vesicles still formed? Is the cargo still loaded on vesicles? One has pretty much the tendency to put together biogenesis/release but the mechanisms involved in the generation and then in the release are very distinct. The manuscript should be clear on this point.

6) In subsection “Wnt signaling in EV biology” the authors test the effect of Wnt3a on EV release and show a decrease (is it on biogenesis or the release? by which mechanism?). This appears different from what was previously described in other cells such as microglia (Hooper et al., 2012), where Wnt3a induces exosome release, and inhibition of GSK3 does not altered exosome secretion, indicating that exosome release occurs through a GSK3-independent mechanism. Such discrepancy is not described, and the paper not quoted if I am not wrong.

7) In concern to the effects of YKT6 that has been described by the group of Boutros as required for exosome-dependent Wnt release in Drosophila wing disc and cells and Hela cells, the paper of Gross et al., 2012 is not quoted.

8) In Figure 1—figure supplement 1B and in the legend there is a discrepancy: in the figure it is 12.200g and in the legend 10.000 g and the time of not indicated in the figure.

9) Figure 5B in the text there is a reference to cells versus EVs (subsection “A CRISPR/Cas9 screen using bEXOmiRs”) though unclear in the Figure.

In conclusion, it is an elegant set up that deserves attention but certainly additional controls. At this point in my opinion it is more the set-up of a new method that maybe will set the stage for more meaningful experiments in the future. This may help to better target potential mechanisms and therefore understand the crosstalks and pathways influencing EV (rather subpopulations of EVs) biogenesis and release which should be separated as this are very different processes.

Reviewer #3:

In this manuscript, Albert Lu and colleagues, have designed a screen to identify regulators of the secretion of extracellular vesicles (EV). They have used a clever design by combining bar-coding of the EV with a single guide RNA to target genes using the Cas9 enzyme. Bar-coding is achieved using custom-designed miRNAs, the bEXOmiR.

The authors used a pool screen format where an excess of cells is transfected with a complex library of viruses. Next generation sequencing then quantifies the abundance of specific miRNA in the supernatant of the pool of cells. Because each gene-specific guide RNA is combined with a specific bEXOmiR, inference can be drawn from the levels of a bEXOmiR in the supernatant.

This is an elegant study that provides a valuable resource for researchers interested in EV. It also provides a proof of concept that such bar-coding of EV is feasible. Finally, it demonstrates elegantly that Wnt signaling is involved in regulating EV secretion.

In such pooled screens, the normalisation steps are really key to avoid mis-interpreting the data. In this case, Lu and colleagues have normalised first the bEXOmiR count to a cell line devoid of the nuclease Cas9. This should exclude any effect that the bEXOmiR barcode could have on packaging into EV or the effect the miRNAs could have on gene expression. The authors show convincingly that this is the case.

The next concern would be that the CRISPR targeted gene would affect the synthesis of the bar-coded miRNA instead of packaging and export in EV. These are admittedly two very different processes and it would be advantageous to distinguish them.

As far as I could see, the authors did not use the comparison extra/intra cellular levels of bEXOmiR to sort their hits. It would be interesting to sort the hits based on whether there is an individual correlation between intracellular and extracellular amount of bEXOmiR upon gene knockout. This should be the case for the hits proposed to be involved in "miRNA processing and trafficking" (proposal based on previous annotations) (Figure 4). In theory it could help sort the other hits between miRNA processing and EV biosynthesis, especially for RNA binding proteins.

The final concern with screen design would be whether some genes could affect the uptake and degradation of EV. Indeed, as this is a suspension culture, cells have in theory the capacity to affect the extracellular levels of EV also through endocytosis and degradation. A hit that would affect the membrane surface of EV and thus affect EV binding to neighbouring cells after secretion could, in theory affect the overall levels of EV without affecting their synthesis and release. I wonder if this could be an explanation for the effect of some of the extracellular proteins identified (Figure 4). I wonder if the authors can exclude this effect or control for it?

" Unexpectedly, YKT6, which appeared as a suppressor in the initial screen (Figure 2B), was uncovered as a slight activator instead; this may be related to differences in the bEXOmiRs associated with this gene in the two separate analyses..… This discrepancy highlights the importance of orthogonal hit validation experiments."

It also highlights the difficulty in interpreting the significance score (-Log10(P-value) to evaluate reproducibility. I have trouble evaluating how many pairs of bEXOmiRs-sgRNA gave similar results with this metric.

It would be surprising that the YKT6 case would occur if the result found in the main screen is attributable to 3 or 4 pairs of bEXOmiR+sgRNA. I think it would help clarify the data if the authors provided a table with the result for each gene: with how many constructs gave a result and what score.

As you will see from the reviews, the reviewers' opinions of your study were mixed. All reviewers feel that the bEXOmiR experimental approach is clever and may hold promise for identifying EV biogenesis pathway components. However, while the reviewers acknowledge the potential trove of new information regarding EV biogenesis, they found the breadth and diversity of the 'hits' a matter of some concern, because for most cases it is not possible to distinguish "fundamental" versus "peripheral" hits in EV biogenesis pathways. It is a fair assumption that many of the hits impact EV biogenesis and release due to general perturbations to organelle function (that is, they do not directly participate in some aspect of EV biology). The consensus view is that additional experimentation is necessary to sufficiently validate the bEXOmiR approach and the screening data, and the reviewers request that you outline additional experimentation to address this. The reviewers would be especially encouraged by experimental data regarding mechanism: what aspect of EV biogenesis is affected by loss of ARHGEF18 and/or is regulated by Wnt signaling? The reviewers felt that the Wnt signaling and ARHGEF18 follow-up work was largely confirmatory of the original screen hits but did not provide strong additional evidence for a fundamental role of these factors in EV biogenesis or regulation. The reviewers would also have liked to see some additional data that would provide insight into the origin of the bEXOmiR-containing EV vesicle (plasma membrane? endosome? another organelle?).

We believe that the live cell imaging shows that loss of EVs from our studied hits is due to decreased exocytosis of multivesicular endosomes – a very important new finding. In addition, we provide a molecular explanation for how the GSK3 inhibition blocks K562 MVE release – by decreasing Rab27 mRNA and protein levels.

Reviewer #1:

The manuscript describes the construction and validation of a genetic screening method that links specific sgRNA sequences with synthetic miRNAs that are secreted via extra-cellular vesicles (EVs). The method allows one to use deep sequencing to gauge the effects of targeted gene knockouts on EV-mediated release of miRNAs on a whole genome scale. The method was applied in replicate screens, and its effectiveness is suggested by the identification of a handful of gene products previously implicated in EV biogenesis and/or release. In addition, at least one gene, ARHGEF18, was shown to be required for EV-mediated release of endogenous miRNAs.

I find the method to be a clever implementation of miRNA and sgRNA processing pathways to link genetics with phenotype. The most important question is, does it work?

Thank you.

The identification of gene products previously identified as required for EV release suggests that it does.

We agree.

The approach did, however, require the authors to tackle a few technical bugs (eg, DNA sequencing, effects of different sgRNA-barcode combinations, etc) and it is not clear if further, unknown biases (if any) impact the utility of the method as a general method for evaluating the genetics of EV biogenesis.

We are very honest about limitations of the screen and explored some of them – it is genome wide, but we do not claim that it is a saturation screen.

A large number of genes not previously implicated in EV biology are asserted to play roles in EV biogenesis. At this stage, it is unclear if any of the 'new' hits play a direct role in EV cargo sorting, EV biogenesis, or EV release.

Our live cell pHluorin-CD63 TIRF experiments strongly suggest that MVE release is the problem in characterized hit-depleted cells – CD63-positive structures don’t appear to change, but their probability of release does.

As with most screens, the ultimate utility of the data will require substantial amount of follow-up work and will not be known for years to come.

We agree – thanks for acknowledging this.

Reviewer #2:

In this manuscript Lu and colleagues perform a genome wide screen to get insights into the biogenesis and regulation of Extracellular Vesicle release.

For this purpose, the authors set up an elegant system based on the generation of barcoded miRNAs that depict a sequence that has been previously shown to target miRNAs to multivesicular endosomes of T lymphocytes and in particular to their intralumenal vesicles (called exosomes once they are secreted). These findings from the group of Sanchez Madrid are largely exploited here (and would deserve better acknowledgement).

We did cite their paper and added additional acknowledgement.

The barcoded miRNAs are then used as reporters of EV release which is followed by a genome wide screen coupled with CRISPR/Cas9.

The authors have performed an incredible amount of work, that certainly is a very interesting set up.

There are several concerns with the manuscript, interpretation of results and the claims. Finally, one is a little left with the question on how such approach can really answer fundamental questions and pinpoint general or particular/specific machineries required for EV biogenesis and release.

Essential revisions:

1) EVs comprise very different types of vesicles that are either endosome derived (so called exosomes) and plasma membrane derived. One important issue in the field is the possibility in the future to ascribe functions to the different subpopulations. Several studies have described and reported distinct machineries for the biogenesis and release of exosomes and microvesicles that can be cell type and cargo specific (summarized in different recent reviews from different authors). Therefore, it is unclear what is the approach that is here developed bringing to the complexity that has been observed and described for the very different systems.

We have added new NTA Nanoparticle tracking which shows that our vesicles are rather uniform and small in size and more importantly, we added live cell imaging that now shows that our hits are really important for MVE exocytosis. These are very major new additions that tell us about the pathway involved, greatly increasing the significance of our story.

2) It is unclear from the claims in the manuscript and from the experiments done if one is trying to get hints on the biogenesis and release of endosome-derived exosomes or plasma membrane derived vesicles. It would certainly be important for the general audience to be clearer on this issue that is a real concern.

See above – our new data address this very well.

With this in mind, the characterization made, that is far from being very extensive, would seem to indicate that the authors maybe rather focus on endosome derived exosomes. This point should be consolidated better with additional proteins tested by Western blot, CD63, Tsg101. For example, CD63 should rather be used for the immunofluorescence experiments presented. Lamp1, if present in late endosomes, is much more enriched in lysosomes. The experiments presented in Figure 1 and Figure 6 are difficult then to interpret.

We have used CD63 is our exocytosis marker and we also now include CD63 IF (new Figure 1—figure supplement 2, Figure 5 and Figure 5—figure supplement 3.

Lamp1 is far from being the best "exosome" marker as in most cells Lamp1 is on the limiting membrane of the organelle and it is Lamp2 that is rather on the intraluminal vesicles. Therefore, it would be important to control whether in the cells under study Lamp1 is targeted to the intraluminal vesicles (or even if this is a consequence of the barcoded miRNAs?)

We added additional markers thus this is not needed at this stage (see above).

3) Following point 2, electron microscopy combined with extra immunofluorescence imaging would be more then appropriate and required to be sure, even to set background for the all findings, that the barcoded miRNAs, once expressed are not impacting in cell/organelle morphology and/or if they are targeted to the right place. This could also help understanding the effects of Rab27A which in most cells where it is expressed (not all cells express the "a" isoform) is present in Golgi TGN. Rab27a is localized to Lysosome Related organelles only in specialized cells. Therefore, is Rab27a in these manipulated cells recruited to multivesicular endosomes that may specialize after introduction of the barcoded miRs? For example, this is the case in HeLa cells expressing MHC Class II and all associated molecules for antigen presentation (Ostrowski et al., 2010). The sequence used has been described to my knowledge only in T cells, is it really driving the miRs to the right place in very different cells? In conclusion extra controls would support part of the claims. This would then help understanding what subpopulation of EVs is targeted in the study.

The live cell imaging really supports the idea that MVE release is what is impacted here. The NTA shows it is small vesicles. We have added more IF to show that the bEXOmiRs don’t change CD63compartment morphology (new Figure 1—figure supplement 2). As reported by others and shown here, Rab27 depletion decreases MVE exocytosis and EV recovery.

4) I would be more careful generalizing the concept based on the two cell lines used in this study although both express wnt.

We have modified the text as requested and toned down the Title.

5) The manuscript describes the involvement of Wnt signaling in the regulation of « EV biogenesis ». For me it is unclear what is really the mechanism of Wnt action: is it on the biogenesis per se or on the release/secretion step which are different parts of the pathway? It could be only on the release but whether there are biogenetic functions remains unclear. Are vesicles still formed? Is the cargo still loaded on vesicles? One has pretty much the tendency to put together biogenesis/release but the mechanisms involved in the generation and then in the release are very distinct. The manuscript should be clear on this point.

Again, we now use live cell imaging and show that loss of Rab27 blocks EV exocytosis and importantly, that GSK3 inhibition decreases Rab27 mRNA and protein. Thus, the pathway is now much clearer.

6) In subsection “Wnt signaling in EV biology” the authors test the effect of Wnt3a on EV release and show a decrease (is it on biogenesis or the release? by which mechanism?). This appears different from what was previously described in other cells such as microglia (Hooper et al., 2012), where Wnt3a induces exosome release, and inhibition of GSK3 does not altered exosome secretion, indicating that exosome release occurs through a GSK3-independent mechanism. Such discrepancy is not described, and the paper not quoted if I am not wrong.

We confirm here that different cells have different responses to GSK3 inhibition and have included this reference.

7) In concern to the effects of YKT6 that has been described by the group of Boutros as required for exosome-dependent Wnt release in Drosophila wing disc and cells and Hela cells, the paper of Gross et al., 2012 is not quoted.

We apologize and added this reference.

8) In Figure 1—figure supplement 1B and in the legend there is a discrepancy: in the figure it is 12.200g and in the legend 10.000 g and the time of not indicated in the figure.

Thank you for detecting that error-now corrected.

9) Figure 5B in the text there is a reference to cells versus EVs (subsection “A CRISPR/Cas9 screen using bEXOmiRs”) though unclear in the Figure.

We fixed this figure – they are now labeled more clearly at the top.

In conclusion, it is an elegant set up that deserves attention but certainly additional controls. At this point in my opinion it is more the set-up of a new method that maybe will set the stage for more meaningful experiments in the future. This may help to better target potential mechanisms and therefore understand the crosstalks and pathways influencing EV (rather subpopulations of EVs) biogenesis and release which should be separated as this are very different processes.

We thank the reviewer and wish to remind all the reviewers that this was submitted as a Resource.

Reviewer #3:

In this manuscript, Albert Lu and colleagues, have designed a screen to identify regulators of the secretion of extracellular vesicles (EV). They have used a clever design by combining bar-coding of the EV with a single guide RNA to target genes using the Cas9 enzyme. Bar-coding is achieved using custom-designed miRNAs, the bEXOmiR.

The authors used a pool screen format where an excess of cells is transfected with a complex library of viruses. Next generation sequencing then quantifies the abundance of specific miRNA in the supernatant of the pool of cells. Because each gene-specific guide RNA is combined with a specific bEXOmiR, inference can be drawn from the levels of a bEXOmiR in the supernatant.

This is an elegant study that provides a valuable resource for researchers interested in EV. It also provides a proof of concept that such bar-coding of EV is feasible. Finally, it demonstrates elegantly that Wnt signaling is involved in regulating EV secretion.

Thank you.

In such pooled screens, the normalisation steps are really key to avoid mis-interpreting the data. In this case, Lu and colleagues have normalised first the bEXOmiR count to a cell line devoid of the nuclease Cas9. This should exclude any effect that the bEXOmiR barcode could have on packaging into EV or the effect the miRNAs could have on gene expression. The authors show convincingly that this is the case.

Thank you.

The next concern would be that the CRISPR targeted gene would affect the synthesis of the bar-coded miRNA instead of packaging and export in EV. These are admittedly two very different processes and it would be advantageous to distinguish them.

As far as I could see, the authors did not use the comparison extra/intra cellular levels of bEXOmiR to sort their hits. It would be interesting to sort the hits based on whether there is an individual correlation between intracellular and extracellular amount of bEXOmiR upon gene knockout. This should be the case for the hits proposed to be involved in "miRNA processing and trafficking" (proposal based on previous annotations) (Figure 4). In theory it could help sort the other hits between miRNA processing and EV biosynthesis, especially for RNA binding proteins.

We have not sorted our hits according to packaging efficiency and note this in the revised text. We did address this in Figure 5 for the Nanostring experiment where specific miRNA levels only change in the EV but not cellular fractions. Orthogonal validation is thus important to ensure it was not a bEXOmiR synthesis issue - monitoring the release of endogenous miRNAs instead of bEXOmiRs and creating CRISPR cell lines that show the same phenotype for specific hits without any bEXOmiRs. By using 10 different bEXOmiRs for every gene, we tried to minimize any biosynthesis bias. In addition, we now more clearly explain Figure 3 where we show no correlation between abundance of total bEXOmiRs in cells compared with EVs. We also validated the roles of the hits using TIRF, confirming the use of these reporters to good end.

The final concern with screen design would be whether some genes could affect the uptake and degradation of EV. Indeed, as this is a suspension culture, cells have in theory the capacity to affect the extracellular levels of EV also through endocytosis and degradation. A hit that would affect the membrane surface of EV and thus affect EV binding to neighbouring cells after secretion could, in theory affect the overall levels of EV without affecting their synthesis and release. I wonder if this could be an explanation for the effect of some of the extracellular proteins identified (Figure 4). I wonder if the authors can exclude this effect or control for it?

Although endocytosis of EVs by surface proteins might lead to a decrease in EV bEXOmiR abundance, we did not detect a large cluster of endocytosis genes in our screen and state this in the text.

" Unexpectedly, YKT6, which appeared as a suppressor in the initial screen (Figure 2B), was uncovered as a slight activator instead; this may be related to differences in the bEXOmiRs associated with this gene in the two separate analyses..… This discrepancy highlights the importance of orthogonal hit validation experiments."

It also highlights the difficulty in interpreting the significance score (-Log10(P-value) to evaluate reproducibility. I have trouble evaluating how many pairs of bEXOmiRs-sgRNA gave similar results with this metric.

Our screen has many limitations and the CASTLE score emphasizes the outliers. Again, different bEXOmiRs have variable efficiencies as do different guides, and one would need to redo the screen with a more limited, enhanced set to have better genome-wide coverage. We have added better data on recovery of each bEXOmiR for all the hits in a new Table 7 as requested.

It would be surprising that the YKT6 case would occur if the result found in the main screen is attributable to 3 or 4 pairs of bEXOmiR+sgRNA. I think it would help clarify the data if the authors provided a table with the result for each gene: with how many constructs gave a result and what score.

Done as requested – usually 2-3 guide-bEXOmiR associations supported each hit. Text modified.

Author details

1. Albert Lu

   Department of Biochemistry, Stanford University School of Medicine, Stanford, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing

   For correspondence

   alulopez@stanford.edu

   Competing interests

   No competing interests declared

2. Paulina Wawro

   Department of Biochemistry, Stanford University School of Medicine, Stanford, United States

   Contribution

   Investigation, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

3. David W Morgens

   Department of Genetics, Stanford University School of Medicine, Stanford, United States

   Contribution

   Data curation, Software, Formal analysis, Validation, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

4. Fernando Portela

   Department of Biochemistry, Stanford University School of Medicine, Stanford, United States

   Contribution

   Data curation, Software, Methodology, Writing—original draft, Writing—review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-0238-9251

5. Michael C Bassik

   Department of Genetics, Stanford University School of Medicine, Stanford, United States

   Contribution

   Conceptualization, Resources, Data curation, Software, Funding acquisition, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

6. Suzanne R Pfeffer

   Department of Biochemistry, Stanford University School of Medicine, Stanford, United States

   Contribution

   Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Methodology, Writing—original draft, Project administration, Writing—review and editing

   For correspondence

   pfeffer@stanford.edu

   Competing interests

   Reviewing editor, eLife

   "This ORCID iD identifies the author of this article:" 0000-0002-6462-984X

• 5,958

  Page views

• 1,025

  Downloads

• 17

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.